RESEARCH. Drug eluting and bare metal stents in people with and without diabetes: collaborative network meta-analysis

Transcription

1 Open Access: free full text available online For the full versions of these articles see bmj.com Drug eluting and bare metal stents in people with and without diabetes: collaborative network meta-analysis Christoph Stettler, 1,2,3 Sabin Allemann, 1,2 Simon Wandel, 1 Adnan Kastrati, 4 MarieClaudeMorice, 5 Albert Schömig, 4 Matthias E Pfisterer, 6 Gregg W Stone, 7 Martin B Leon, 7 José Suárez de Lezo, 8 Jean-Jacques Goy, 9 Seung-Jung Park, 10 Manel Sabaté, 11 Maarten J Suttorp, 12 Henning Kelbaek, 13 Christian Spaulding, 14 Maurizio Menichelli, 15 Paul Vermeersch, 16 Maurits T Dirksen, 17 Pavel Cervinka, 18 Marco De Carlo, 19 Andrejs Erglis, 20 Tania Chechi, 21 Paolo Ortolani, 22 Martin J Schalij, 23 Peter Diem, 2 Bernhard Meier, 24 Stephan Windecker, 24,25 Peter Jüni 1,25 EDITORIAL by Agostini and Van den Branden 1 Institute of Social and Preventive Medicine, University of Bern, 3012 Bern, Switzerland 2 Division of Endocrinology, Diabetes and Clinical Nutrition, University Hospital, Bern, Switzerland 3 International Centre for Circulatory Health, National Heart and Lung Institute, Imperial College, London 4 Deutsches Herzzentrum, Technische Universität, Munich, Germany 5 Institut Hospitalier Jacques Cartier, Massy, France 6 Division of Cardiology, University Hospital, Basel, Switzerland 7 Columbia University Medical Center, New York, USA 8 Servicio de Cardiología, Hospital Universitario Reina Sofía, Cordoba, Spain 9 Service of Cardiology, Clinique Cecil, Lausanne, Switzerland 10 Department of Medicine, ASAN Medical Center, Seoul, Korea 11 Department of Interventional Cardiology,HospitaldeSanPau, Barcelona, Spain 12 Department of Interventional Cardiology, St Antonius Hospital, Nieuwegein, Netherlands 13 Cardiac Catheterisation Laboratory, Rigshospitalet, Copenhagen, Denmark 14 Cochin Hospital, Assistance Publique Hôpitaux de Paris, Paris Descartes University, and INSERM U 780, Paris, France 15 San Camillo Hospital, Rome, Italy 16 Antwerp Cardiovascular Institute Middelheim, Antwerp, Belgium ABSTRACT Objective To compare the effectiveness and safety of three types of stents (sirolimus eluting, paclitaxel eluting, and bare metal) in people with and without diabetes mellitus. Design Collaborative network meta-analysis. Data sources Electronic databases (Medline, Embase, the Cochrane Central Register of Controlled Trials), relevant websites, reference lists, conference abstracts, reviews, book chapters, and proceedings of advisory panels for the US Food and Drug Administration. Manufacturers and trialists provided additional data. Review methods Network meta-analysis with a mixed treatment comparison method to combine direct within trial comparisons between stents with indirect evidence from other trials while maintaining randomisation. Overall mortality was the primary safety end point, target lesion revascularisation the effectiveness end point. Results 35 trials in 3852 people with diabetes and people without diabetes contributed to the analyses. Inconsistency of the network was substantial for overall mortality in people with diabetes and seemed to be related to the duration of dual antiplatelet therapy (P value for interaction 0.02). Restricting the analysis to trials with a duration of dual antiplatelet therapy of six months or more, inconsistency was reduced considerably and hazard ratios for overall mortality were near one for all comparisons in people with diabetes: sirolimus eluting stents compared with bare metal stents 0.88 (95% credibility interval 0.55 to 1.30), paclitaxel eluting stents compared with bare metal stents 0.91 (0.60 to 1.38), and sirolimus eluting stents compared with paclitaxel eluting stents 0.95 (0.63 to 1.43). In people without diabetes, hazard ratios were unaffected by the restriction. Both drug eluting stents were associated with a decrease in revascularisation rates compared with bare metal stents in people both with and without diabetes. Conclusion In trials that specified a duration of dual antiplatelet therapy of six months or more after stent implantation, drug eluting stents seemed safe and effective in people both with and without diabetes. INTRODUCTION People with diabetes are at an increased for coronary heart disease and have more restenoses after the implantation of coronary stents. Sirolimus eluting and paclitaxel eluting stents are associated with a noticeable reduction in target lesion revascularisation compared with bare metal stents, whereas the rates of overall mortality and cardiac mortality are similar. 1 Randomised trials have reported a reduced revascularisation rate with sirolimus eluting and paclitaxel eluting stents than with bare metal stents in people with diabetes, 2-4w1-w3 but a subsequent meta-analysis of four early trials in 428 people with diabetes that compared sirolimus eluting stents with bare metal stents for up to four years suggested a strongly increased of mortality (hazard ratio 2.90, 95% confidence interval 1.38 to 6.10), 5 which translated into a number needed to harm to cause one death over four years as low as 4 (95% confidence interval 2 to 22). In a previous network meta-analysis 1 we determined the average benefits and harms of all three stents and provided preliminary results for overall mortality and the composite of death or myocardial infarction stratified according to diabetes status. Here we extend that analysis. 67 We prespecified overall mortality as the primary safety outcome and explored the consistency of mortality data in people with diabetes. METHODS We included randomised controlled trials in people with symptoms or signs of myocardial ischaemia that compared two first generation drug eluting stents (containing paclitaxel or sirolimus) with each other or with a bare metal stent. Trials had to have a clinical follow-up of at least six months. We specified overall mortality as the primary safety outcome and target lesion revascularisation as the 668 BMJ 20 SEPTEMBER 2008 VOLUME 337

2 17 Amsterdam Department of Interventional Cardiology, Onze LieveVrouweGasthuis, Amsterdam, Netherlands 18 Division of Cardiology, Masaryk Hospital, Ústí nad Labem, Czech Republic 19 Cardiothoracic and Vascular Department, University of Pisa, Italy 20 University of Latvia, Institute of Cardiology, Riga, Latvia 21 Cardiologia e Cardiologia Invasiva 2 AOU-Careggi, Florence, Italy 22 Institute of Cardiology, S Orsola- Malpighi Hospital, University of Bologna, Italy 23 Department of Cardiology, Leiden University Medical Center, Leiden, Netherlands 24 Department of Cardiology, Bern University Hospital, Bern, Switzerland 25 CTU Bern, Bern University Hospital, 3010 Bern, Switzerland Correspondence to: P Jüni juni@ispm.unibe.ch Cite this as: BMJ 2008;337:a1331 This article is an abridged version of a paper that was published on bmj.com. Cite this article as: BMJ 2008;337:a1331 primary effectiveness outcome. Secondary safety outcomes were cardiac death, a composite of death or myocardial infarction, and stent thrombosis, in accordance with criteria of the Academic Research Consortium. We also recorded stent thrombosis according to the definitions used in individual trials (per protocol definition). The numbers of patients experiencing an event and the overall number of patients at were recorded separately for years 1to4. To tackle potential sources of inconsistency we determined the month of completed patient recruitmentasameasureofthetimeperiodwhenatrialwas done, the duration of follow-up (years), and the duration (months) of dual antiplatelet therapy and assessed concealment of allocation, blinding of staff adjudicating clinical outcomes, and adherence to the intention to treat principle, with trials considered to be of high quality that satisfied all three criteria. Statistical analysis We used a hierarchical random effects model 8 for mixed treatment comparisons 9 basedonpiecewise constant hazards, with random effects at the levels of trials, adjacent time periods, and comparisons. We simultaneously estimated log hazard ratios and the difference in log ratios for people with and without diabetes. From the posterior distribution of the difference we estimated the P values for interaction between treatment effect and diabetes status. Hazard ratios were estimated from the median, and the accompanying 95% credibility intervals from the 2.5th and 97.5th centiles of the posterior distribution. As event rates were low we derived relative s of stent thromboses using a random effects Poisson regression model. 10 We did separate analyses according to time of occurrence of stent thrombosis and an analysis of per protocol definitions of stent thrombosis. We evaluated the inconsistency expressed as the percentage difference in hazard ratios between direct comparisons within trials and indirect comparisons between trials. Values of 0% indicate no inconsistency of comparisons in the network, 100% indicate high inconsistency (see bmj.com)..we evaluated heterogeneity between trials, defined as variability of results across trials within comparisons over and above chance, and the model s goodness of fit to the data. To investigate potential sources of variation in the network, we included the characteristics of the trials as covariates in the network meta-analysis of the primary safety outcome. We used prespecified cut-off points of two years for follow-up, January 2004 for completed patient recruitment, and six months for dual antiplatelet therapy. We used a random effects Poisson regression model for analyses where the number of trials and events was too low to allow the estimation of random effects at the level of time periods. 10 Pvalues for interaction between trial characteristics and treatment effect were derived from the posterior distribution of covariates. Duration of dual antiplatelet therapy was the only variable with an interaction with P<0.05. Therefore we restricted the dataset to trials with a duration of dual antiplatelet therapy of six months or longer and repeated all analyses. RESULTS w1-w14 w16-w33 w35 w38 w40 Thirty five trials were included (3852 with diabetes and without diabetes). w1 w4-w14 w16- Additional data were provided for 32 trials. w19 w21-w33 w35 w38 w40 Recruitment was completed between January 2001 w16 and November w5 Data stratified according to diabetes status were available for all trials on all outcomes, except stent thrombosis. Twenty nine trials reported appropriate w1 w2 w7 w8 w11-w14 w16-w33 w35 w38 w40 allocation concealment, and 28 trials reported blind adjudication for clinical w1 w4 w7 w9-w14 w16-w26 w28-w30 w32 w33 w35 w38 w40 outcomes. For 30 trials patients could be included according to the w1 w3-w7 w9 w11 w13 w14 w16-w23 w25- intention to treat principle. w33 w35 w38 w40 Twenty four trials were of high quality. w1 w7 w11 w13 w14 w16-w23 w25 w26 w28-w30 w32 w33 w35 w38 w40 Duration of dual antiplatelet therapy Study protocols specified the duration of dual antiplatelet therapy as two months in five, w16 w18-w20 w26 three months in three, w3 w7 w17 w2 w5 w9- six months in 18, w14w21w22w24w28w29w32w33w35w38w40 nine months in one, w4 w1 w6 w8 w23 w25 w27 and12monthsineighttrials. w30 w31 All eight trials with dual antiplatelet therapy lasting less than six months compared sirolimus eluting stents with bare metal stents. Auxiliary information indicated that the percentage of patients actually receiving therapy of six months or more was likely to be below 10% in five trials, w7 w16-w19 and for the other three trials between 10% and 20%, w20 40% and 50% w26 and 50% and 60%. w3 One head to head comparison had specified a minimal duration of therapy for two months with sirolimus stents and six months with paclitaxel stents, but the actual duration was only one month shorter for patients receiving siroliumus stents than for those receiving paclitaxel stents, and about 50% of patients allocated to either stent were still receiving therapy after eight months. w32 Network of all trials: overall mortality In people with diabetes the estimated hazard ratio was 1.14 for sirolimus compared with bare metal stents (95% credibility interval 0.74 to 1.60), 1.09 for paclitaxel versus bare metal stents (0.71 to 1.66), and 1.02 for sirolimus versus paclitaxel stents (0.70 to 1.57), compatible with both substantial harm and moderate benefit of either drug eluting stent compared with bare metal stents (see bmj.com). The BMJ 20 SEPTEMBER 2008 VOLUME

3 corresponding values for people without diabetes were 1.02 (0.77 to 1.29), 0.90 (0.67 to 1.16), and 1.13 (0.83 to 1.54; see bmj.com). A moderate to high inconsistency of 61% was found among people with diabetes, but none among people without diabetes. Exploration of sources of variation Estimates of relative comparing sirolimus stents with paclitaxel stents depended to some extent on the quality of the trials, length of follow-up, and time of completed recruitment (see bmj.com), but 95% credibility intervals were wide and tests for interaction negative (P 0.16). The estimated relative of death when sirolimus stents were compared with bare metal stents was greater when dual antiplatelet therapy lasted less than six months (2.37, 95% credibility interval 1.18 to 5.12) compared with six months or longer (0.89, 0.58 to 1.40, P for interaction 0.02), however. When three trials originally classified to have a short duration of dual antiplatelet therapy w3 w20 w26 were reclassified to have a duration of six months or longer, reflecting the auxiliary information indicating that more than 10% of patients in these trials were still receiving therapy at six months, differences were maintained. Restricted network: overall mortality When the network was restricted to trials with dual antiplatelet therapy of six months or longer (see bmj.com), the hazard ratios of death overall among people with diabetes were below 1: sirolimus stents compared with bare metal stents 0.88 (95% credibility interval 0.55 to 1.30), paclitaxel stents compared with bare metal stents 0.91 (0.60 to 1.38), and sirolimus compared with paclitaxel stents 0.95 (0.63 to 1.43). Compared with the network of all trials the inconsistency decreased to 20% and credibility intervals of hazard ratios became more narrow in the restricted network. Among people without diabetes results were much the same in the overall and the restricted network (see bmj.com). The hazard ratio was 1.05 for sirolimus stents compared with bare metal stents (0.69 to 1.73), 0.89 for paclitaxel stents compared with bare metal stents (0.66 to 1.18), and 1.23 for sirolimus compared with paclitaxel stents (0.82 to 1.69). See figures on bmj.com for cumulative incidences of death for the three stent types estimated from the restricted network meta-analysis for people with and without diabetes. The incidence of death was about twice as high in people with diabetes compared with people without diabetes. Tests for interaction between treatment effect and diabetes status were negative for all comparisons (P 0.28). Restricted network: secondary safety outcomes Among people with diabetes, hazard ratios for drug eluting stents compared with bare metal stents became more beneficial for drug eluting stents for the outcomes of cardiac death, the composite of death or myocardial infarction, and stent thromboses in the restricted network compared with the network of all trials (see bmj.com). The inconsistency decreased mainly for cardiac death and per protocol definitions of stent thromboses. No differences between overall and restricted network meta-analysis were observed for myocardial infarction. Among people without diabetes, results from overall and restricted network meta-analysis were similar. Corresponding cumulative incidences for the three stent types are on bmj.com. Again, incidences were higher in people with diabetes than without, with most pronounced differences observed for cardiac death. Tests for interaction between treatment effect and diabetes statuswerenegativeforallcomparisonsoncardiac death, myocardial infarction, and their composite (P 0.47). Among people with diabetes little evidence was found for an increased of definite or per protocol stent thrombosis associated with sirolimus stents compared with either of the two other stents; all point estimates were below 1 and differences in favour of sirolimus stents became more pronounced with the use of per protocol definitions (see bmj.com). For the comparison of paclitaxel stents with bare metal stents all estimates were imprecise for both the Academic Research Consortium definition of definite stent thrombosis and the per protocol definitions. Among people without diabetes relative s were generally higher for both definitions, but tests for interaction between treatment effect and diabetes status were positive only for sirolimus compared with bare metal stents on per protocol definitions of stent thrombosis between day 0 and 4 years and between day 30 and 4 years (P for interaction=0.01). Restricted network: target lesion revascularisation Both drug eluting stents were robustly associated with a decrease in revascularisation rates compared with bare metal stents, network inconsistency was low, and results were unaffected by the restriction of the analysis to trials with dual antiplatelet therapy of six months or more in people with and without diabetes (see bmj.com). Differences between sirolimus and paclitaxel stents tended to be less pronounced among people with diabetes, and credibility intervals overlapped 1. Tests of interaction between treatment effect and diabetes status were negative (P 0.44). Between trial heterogeneity, model fit, and comparison with conventional meta-analyses After restricting the network to trials with dual antiplatelet therapy of six months or more, heterogeneity between trials was low for all outcomes except stent thrombosis. For people without diabetes heterogeneity between trials was low for death, myocardial infarction, and their composite, and moderate for 670 BMJ 20 SEPTEMBER 2008 VOLUME 337

4 WHAT IS ALREADY KNOWN ON THIS TOPIC People with diabetes are at an increased for coronary heart disease and have more restenoses after the implantation of coronary stents A meta-analysis suggested a strongly increased of death associated with sirolimus eluting stents compared with bare metal stents in people with diabetes WHAT THIS STUDY ADDS Reported increases in the of death associated with drug eluting stents compared with bare metal stents in people with diabetes were probably due to dual antiplatelet therapy lasting less than six months in early trials In trials with dual antiplatelet therapy for six months or longer drug eluting stents were safe and effective in people both with and without diabetes In clinical practice it seems prudent to adhere to a minimal duration of dual antiplatelet therapy of six months in all patients undergoing implantation of a drug eluting coronary stent most comparisons on the other outcomes. The model fit was adequate for all outcomes. DISCUSSION Our collaborative network meta-analysis suggests that previously reported increases in the of death associated with sirolimus eluting stents compared with bare metal stents in people with diabetes 5 probably resulted from the restricted duration of dual antiplatelet therapy of less than six months in early trials. In trials with therapy of less than six months, the of death associated with sirolimus stents was more than twice that associated with bare metal stents that is, a number needed to harm as low as 7 to cause one death over four years. Conversely, trials with therapy of six months or more showed no increase in from using sirolimus stents compared with bare metal stents. Restricting the network to trials with therapy of six months or longer resulted in a clear reduction of the inconsistency and hazard ratios near 1, which were robust to all sensitivity analyses. We found similar patterns for analyses of cardiac death and stent thromboses. Compared with bare metal stents, target lesion revascularisation rates are strongly decreased by the use of sirolimus and paclitaxel eluting stents in people with and without diabetes. Numbers needed to treat to reduce one event over four years are 6 in people with diabetes and 8 in people without diabetes. Numbers needed to treat assuming revascularisation rates as observed in routine clinical practice without active angiographic follow-up 11 were estimated 13 for people with diabetes and 18 for people without diabetes. Our study comprises a large body of evidence from randomised controlled trials in people with and without diabetes treated with one of two drug eluting stents or bare metal stents. Additional data were provided according to uniform outcome definitions, including stent thrombosis according to the Academic Research Consortium consensus. 7 This increases comparability between trials and limits bias. Our model was based on relative treatment effects (log hazards ratio). Variations in characteristics of patients or lesions between trials are fully accounted for by maintaining randomised comparisons within each trial. 1 Network meta-analysis makes similar assumptions to standard meta-analysis of direct comparisons within trials, but requires that these assumptions hold over the entire set of trials in the network. 12 The smaller the heterogeneity between trials and the smaller the inconsistency of the data, the more likely relative treatment effects originate from the same distribution and the less likely small trials get undue weight in the analysis. Our exploration of inconsistency is observational in nature and has the same limitations as other observational studies. 13 Most importantly, earlier trials had specified shorter durations of dual antiplatelet therapy than later trials. The duration of therapy was therefore bound to be negatively correlated with the duration of follow-up, and confounding could exist between these variables. Other potential confounders include changes over time in patient selection and procedural characteristics or methodological quality. We addressed this by repeating tests of interaction between treatment effect and components of methodological quality or length of follow-up after the exclusion of trials with dual antiplatelet therapy lasting less than six months and foundnoevidenceforaninteractioninanyofthese analyses (data available on request). Our results could have been corroborated by an analysis of the actual duration of dual antiplatelet therapy in individual patients, but precise durations were unavailable in most trials and we lacked the resources to retrospectively ascertain and validate usage data. Eight trials had specified a duration of therapy of less than six months. For five trials we are confident that the actual duration corresponded to the specified duration in at least 90% of patients w7 w16-w19 ; in three trials between 10% and 50% of the patients w3 w20 w26 had dual antiplatelet therapy at six months. Results were unaffected by the reclassification of these three trials as having a duration of dual antiplatelet therapy of six months or longer, and the P value for interaction between relative of death and duration of therapy became even smaller. Additionally, strut thickness or type of bare metal stent used in comparison groups might affect clinical outcomes. w21 Despite our results being robust to the adjustment for these characteristics, 1 we cannot fully exclude the possibility that differences in bare metal stents as comparators contributed to the observed variation in mortality between trials with short and long durations BMJ 20 SEPTEMBER 2008 VOLUME

5 of therapy. Finally, we were unable to record information on specific antidiabetic treatment or on glycaemic control in people with diabetes mellitus and to perform separate analyses for people with diabetes who did or did not use insulin. Although these aspects are related to cardiovascular outcomes, 14 they were per definition randomly distributed across comparison groups within each trial and it seems unlikely that they influenced results. We found that the duration of dual antiplatelet therapy modified the safety profile of drug eluting stents mainly in people with diabetes. The beneficial effect of prolonged therapy in people with diabetes may be mainly related to differences in lesion characteristics people with diabetes tend to have smaller vessels and longer lesions than people without diabetes. Previous studies have identified these variables as predictors of stent thrombosis, 15 which may explain the predisposition of people with diabetes to stent thrombosis in the absence of adequate antiplatelet therapy. Delays in arterial healing and re-endothelialisation may be more pronounced in people with diabetes, particularly in the absence of dual antiplatelet therapy Finally, people with diabetes may be more likely to experience aspirin resistance than people without diabetes 18 and may have a particular benefit from a complementary antiplatelet therapy with clopidogrel or ticlopidine. Conclusion In trials with a duration of dual antiplatelet therapy of six months or longer drug eluting stents were safe and effective in people with and without diabetes. It seems prudent to adhere to a minimal duration of therapy for six months in patients having drug eluting coronary stents. We thank Boston Scientific and Cordis for the provision of additional data. CTU Bern is supported by the Swiss National Science Foundation. Contributors: See bmj.com. Funding: Swiss National Science Foundation (grant Nos 3233B , , and ) to CSt and PJ. The implementation and validation of the statistical models used for this study were funded by the Swiss National Science Foundation s national research programme 53 (grant No ) to PJ. CSt and PJ are PROSPER (program for social medicine, preventive and epidemiological research) fellows funded by the Swiss National Science Foundation. The funder had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Competing interests: CSt and PJ report receiving unrestricted grants from the Swiss National Science Foundation. AK receives lecture fees from Bristol-Myers Squibb, Cordis, GlaxoSmithKline, Lilly, Medtronic, Novartis, and Sanofi-Aventis. MCM receives lecture fees from Cordis, Boston Scientific, and Abbot, which go to a research organisation (RCF, Massy, France). AS receives unrestricted grant support for the Department of Cardiology he chairs from Amersham/General Electric, Bayerische Forschungsstiftung, Bristol-Myers Squibb, Cordis, Cryocath, Guidant, Medtronic, Nycomed, and Schering. MEP receives lecture fees from Medtronic. GWS receives consulting fees from Boston Scientific, Abbott, Guidant, Xtent, and BMS Imaging, lecture fees from Boston Scientific, Abbott, and Medtronic, has equity interests in Devax and Xtent, and is a member of the board of directors of Devax. MBL receives consulting fees from Cordis, Medtronic, Boston Scientific, and OrbusNeich and has equity interests in Conor, Medinol, and OrbusNeich. JJG is on the advisory board of Boston Scientific and receives research grant support from Cordis. SJP receives research grant support from Cordis. HK receives unrestricted grant support from Cordis. CSp receives consulting and lecture fees from Cordis, Boston Scientific, Abbot, Lilly, and Pfizer. MTD receives lecture fees from Boston Scientific. BM receives research grant support from various stent companies, including Cordis and Boston Scientific, and is in the speaker bureau for various stent companies, including Cordis and Boston Scientific. SW receives lecture and consulting fees from Abbot, Biotronic, Biosensors, Boston Scientific, Cordis, and Medtronic. GWS and MBL are directors of the Cardiovascular Research Foundation, a public charity affiliated with Columbia University Medical Center, from which they receive no compensation; the Cardiovascular Research Foundation receives research or educational funding from Boston Scientific, Cordis, Sanofi-Aventis, and Bristol-Myers Squibb. Ethical approval: Not required. Provenance and peer review: Not commissioned; externally peer reviewed. 1 Stettler C, Wandel S, Allemann S, Kastrati A, Morice MC, Schomig A, et al. Outcomes associated with drug-eluting and bare-metal stents: a collaborative network meta-analysis. Lancet 2007;370: Moussa I, Leon MB, Baim DS, O Neill WW, Popma JJ, Buchbinder M, et al. Impact of sirolimus-eluting stents on outcome in diabetic patients: a SIRIUS (SIRolImUS-coated Bx Velocity balloonexpandable stent in the treatment of patients with de novo coronary artery lesions) substudy. Circulation 2004;109: HermillerJB,RaiznerA,CannonL,GurbelPA,KutcherMA,WongSC, et al. Outcomes with the polymer-based paclitaxel-eluting TAXUS stent in patients with diabetes mellitus: the TAXUS-IV trial. JAmColl Cardiol 2005;45: Abizaid A, Costa MA, Blanchard D, Albertal M, Eltchaninoff H, Guagliumi G, et al. Sirolimus-eluting stents inhibit neointimal hyperplasia in diabetic patients. Insights from the RAVEL trial. Eur Heart J 2004;25: Spaulding C, Daemen J, Boersma E, Cutlip DE, Serruys PW. A pooled analysis of data comparing sirolimus-eluting stents with bare-metal stents. NEnglJMed2007;356: Mauri L, Hsieh WH, Massaro JM, Ho KK, D Agostino R, Cutlip DE. Stent thrombosisinrandomizedclinical trialsofdrug-elutingstents.nenglj Med 2007;356: Cutlip DE, Windecker S, Mehran R, Boam A, Cohen DJ, van Es GA, et al. Clinical end points in coronary stent trials: a case for standardized definitions. Circulation 2007;115: Smith TC, Spiegelhalter DJ, Thomas A. Bayesian approaches to random-effects meta-analysis: a comparative study. Stat Med 1995;14: Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med 2004;23: Cooper NJ, Sutton AJ, Lu G, Khunti K. Mixed comparison of stroke prevention treatments in individuals with nonrheumatic atrial fibrillation. Arch Intern Med 2006;166: Tu JV, BowenJ, Chiu M, KoDT, AustinPC, He Y, etal. Effectivenessand safety of drug-eluting stents in Ontario. NEnglJMed 2007;357: Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ 2005;331: Thompson SG. Why sources of heterogeneity in meta-analysis should be investigated. BMJ 1994;309: Stettler C, Allemann S, Juni P, Cull CA, Holman RR, Egger M, et al. Glycemic control and macrovascular disease in types 1 and 2 diabetes mellitus: meta-analysis of randomized trials. Am Heart J 2006;152: Daemen J, Wenaweser P, Tsuchida K, Abrecht L, Vaina S, Morger C, etal.early andlate coronary stent thrombosisofsirolimus-elutingand paclitaxel-eluting stents in routine clinical practice: data from a large two-institutional cohort study. Lancet 2007;369: Finn AV, Nakazawa G, Joner M, Kolodgie FD, Mont EK, Gold HK, et al. Vascular responses to drug eluting stents: importance of delayed healing. Arterioscler Thromb Vasc Biol 2007;27: JonerM,FinnAV,FarbA,MontEK,KolodgieFD,LadichE,etal. Pathology of drug-eluting stents in humans: delayed healing and late thrombotic. JAmCollCardiol2006;48: Evangelista V, Totani L, Rotondo S, Lorenzet R, Tognoni G, De Berardis G, et al. Prevention of cardiovascular disease in type-2 diabetes: how to improve the clinical efficacy of aspirin. Thromb Haemost 2005;93:8-16. Accepted: 27 June BMJ 20 SEPTEMBER 2008 VOLUME 337

6 Adherence to Mediterranean diet and health status: meta-analysis Francesco Sofi, 1,2,5 Francesca Cesari, 1 Rosanna Abbate, 1,5 Gian Franco Gensini, 3 Alessandro Casini 2,4,5 1 Department of Medical and Surgical Critical Care, Thrombosis Centre, University of Florence, Viale Morgagni 85, Florence, Italy 2 Regional Agency for Nutrition, Azienda Ospedaliero-Universitaria Careggi, Florence 3 Don Carlo Gnocchi Foundation, Onlus IRCCS, Impruneta, Florence 4 Department of Clinical Pathophysiology, Unit of Clinical Nutrition, University of Florence 5 Centro Interdipartimentale per la Ricerca e la Valorizzazione degli Alimenti (CeRA), University of Florence Correspondence to: F Sofi francescosofi@gmail.com Cite this as: BMJ 2008;337:a1344 doi: /bmj.a1344 This article is an abridged version of a paper that was published on bmj.com. Cite this article as: BMJ 2008;337:a1344 ABSTRACT Objective To systematically review all the prospective cohort studies that have analysed the relation between adherence to a Mediterranean diet, mortality, and incidence of chronic diseases in a primary prevention setting. Design Meta-analysis of prospective cohort studies. Data sources English and non-english publications in PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials from 1966 to 30 June Studies reviewed Studies that analysed prospectively the association between adherence to a Mediterranean diet, mortality, and incidence of diseases; 12 studies, with a total of subjects followed for a time ranging from three to 18 years were included. Results The cumulative analysis among eight cohorts ( subjects and deaths) evaluating overall mortality in relation to adherence to a Mediterranean diet showed that a two point increase in the adherence score was significantly associated with a reduced of mortality (pooled relative 0.91, 95% confidence interval 0.89 to 0.94). Likewise, the analyses showed a beneficial role for greater adherence to a Mediterranean diet on cardiovascular mortality (pooled relative 0.91, 0.87 to 0.95), incidence of or mortality from cancer (0.94, 0.92 to 0.96), and incidence of Parkinson s disease and Alzheimer s disease (0.87, 0.80 to 0.96). Conclusions Greater adherence to a Mediterranean diet is associated with a significant improvement in health status, as seen by a significant reduction in overall mortality (9%), mortality from cardiovascular diseases (9%), incidence of or mortality from cancer (6%), and incidence of Parkinson s disease and Alzheimer s disease (13%). These results seem to be clinically relevant for public health, in particular for encouraging a Mediterranean-like dietary pattern for primary prevention of major chronic diseases. INTRODUCTION The Mediterranean diet has been widely reported to be a model of healthy eating for its contribution to a favourable health status and a better quality of life. 12 Since the first data from the seven countries study, 3 several studies in different populations have established a beneficial role for the main components of the Mediterranean diet on the occurrence of cardiovascular diseases and chronic degenerative diseases. 24 However, research interest in this field over the past years has been focused on estimating adherence to the whole Mediterranean diet rather than analysing the individual components of the dietary pattern in relation to the health status of the population. 5 The aim of this study was to systematically review all the available prospective cohort studies in order to establish the role of adherence to a Mediterranean diet in primary prevention. METHODS Study selection We identified studies that prospectively evaluated the association of an a priori score used for assessing adherence to a Mediterranean diet and adverse clinical outcomes (see bmj.com for a description of the search strategy). Our initial search yielded 62 reports, of which 12 articles fulfilled our inclusion criteria. w1-w12 Data extraction We extracted the baseline characteristics from the original reports by using a standardised data extraction form and included them in the meta-analysis. Two investigators (FS and FC) collected the data, and disagreements were solved by consensus and by the opinion of a third author (AC), if necessary. Definition of adherence to Mediterranean diet Adherence to a Mediterranean diet was defined through scores that estimated the conformity of the dietary pattern of the studied population with the traditional Mediterranean dietary pattern. Values of zero or one were assigned to each dietary component by using as cut offs the overall sex specific medians among the study participants. Differences among the studies existed, especially in relation to the food category of vegetables (grouped with potatoes in one study w5 ), meat and meat products (grouped with poultry in some studies w4 w6 ), and nuts w4 w6 w7 w12 and seeds (grouped with fruits in some studies, grouped with legumes in one study, w5 and considered a group by themselves in some others w8 w10 w11 ), as well as milk and dairy products (not present in some studies w8 w10 w11 ) and fish (present only in more recent studies w4-w12 ). Thus, the total adherence scores (estimated as the sum of the above indicated scores of zero and one) varied from a minimum of 0 points indicating low adherence to a maximum of 7-9 points reflecting high adherence to a Mediterranean diet. Statistical analysis We used RevMan, version 4.2 for Windows by the Cochrane Collaboration to analyse data. We used the results of the original studies from multivariable models with the most complete adjustment for potential confounders. We used a random effects model that accounts for interstudy variation and provides a more conservative effect than a fixed model. We calculated random summary relative s with 95% confidence intervals by using an inverse variance method. For details see bmj.com. BMJ 20 SEPTEMBER 2008 VOLUME

7 Study Trichopoulou et al 1995 w1 Kouri-Blazos et al 1999 w2 Lasheras et al 2000 w3 Trichopoulou et al 2003 w4 Knoops et al 2004 w5 Trichopoulou et al 2005 w6 Lagiou et al 2006 w7 Mitrou et al 2007 (men) w11 Mitrou et al 2007 (women) w11 Total Reduced Relative Increased Weight (%) Relative (0.48 to 0.99) (0.50 to 1.25) (0.22 to 1.02) (0.64 to 0.87) (0.82 to 0.94) (0.89 to 0.97) (0.83 to 1.04) (0.91 to 0.94) (0.91 to 0.95) (0.89 to 0.94) Fig 1 Risk of all cause mortality associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size Study Trichopoulou et al 2003 w4 Knoops et al 2004 w5 Mitrou et al 2007 (men) w11 Mitrou et al 2007 (women) w11 Total RESULTS Characteristics of study cohorts Sample sizes varied between 161 and , with a follow-up time ranging from 3.7 to 18 years. Only six out of 12 studies were done in Mediterranean w1 w3-w6 w12 populations. The remaining cohorts comprised US populations, w8-w11 northern Europeans, w5-w7 and a cohort of Europeans living in Australia. w3 Main outcomes Overall mortality was evaluated in eight cohorts (nine studies) for a total of subjects and deaths, cardiovascular mortality in three cohorts (four studies) including a total of subjects and 3876 fatal events, cancer incidence/mortality in five cohorts (six studies) comprising subjects and events, and incidence of Parkinson s disease and Alzheimer s disease in two cohorts (three studies) for a total of subjects and 783 cases. Figure 1 shows the cumulative analysis for studies that analysed overall mortality as the primary clinical outcome. A two point increase in score for adherence to a Mediterranean diet was significantly associated with a reduced of mortality from any cause (relative 0.91, 95% confidence interval 0.89 to 0.94; P<0.0001). Significant heterogeneity was present among the Relative Reduced Increased Weight (%) Relative (0.69 to 0.97) (0.76 to 0.94) (0.89 to 0.96) (0.88 to 0.99) (0.87 to 0.95) Fig 2 Risk of mortality from cardiovascular diseases associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size studies (I 2 =48.8%; P=0.05). However, after exclusion of the paper by Trichopoulou et al 2003 that analysed the same cohort as Trichopoulou et al 2005, w4 w6 the significant association with overall mortality remained (relative 0.92, 0.91 to 0.94; P<0.0001), showing no significant heterogeneity (I 2 =18.3%; P=0.3). Similarly figure 2 shows that a greater adherence to a Mediterranean diet significantly reduced the of mortality from cardiovascular diseases (relative 0.91, 0.87 to 0.95; P<0.0001) with non-significant heterogeneity (I 2 =32.6%; P=0.2). Furthermore, greater adherence to a Mediterranean diet significantly reduced the occurrence of and mortality from neoplasm (relative 0.94, 0.92 to 0.96; P<0.0001) (I 2 =0%; P=0.5) (fig 3). Finally, the overall analysis showed a significant reduction in incidence of Parkinson s disease and Alzheimer s disease associated with a higher score of adherence to a Mediterranean diet (relative 0.87, 0.80 to 0.96; P=0.004), with no heterogeneity among the studies (I 2 =0%; P=0.5) (fig 4). Sensitivity analyses and publication bias Because heterogeneity of studies is likely to produce heterogeneity of effect sizes across studies, we did some sensitivity analyses. These analyses showed no significant influence of any variable (country of origin of the study, sex, follow-up time, quality of the studies) on the overall results of the meta-analysis. We found no evidence of publication bias (see bmj.com). DISCUSSION This meta-analysis, comprising more than 1.5 million healthy subjects and fatal and non-fatal events, shows that greater adherence to a Mediterranean diet is significantly associated with a reduced of overall mortality, cardiovascular mortality, cancer incidence and mortality, and incidence of Parkinson s disease and Alzheimer s disease. The cumulative analysis of 12 cohort studies shows that a two point increase in the score for adherence to a Mediterranean diet determines a 9% reduction in overall mortality, a 9% reduction in mortality from cardiovascular diseases, a 6% reduction in incidence of or mortality from neoplasm, and a 13% reduction in incidence of Parkinson s disease and Alzheimer s disease. To the best of our knowledge, this is the first report that has systematically assessed, through meta-analysis, the possible association between adherence to a Mediterranean diet, mortality, and the occurrence of chronic diseases in the general population. Practical implications We report a significant reduction in of all the main clinical outcomes with an increasing score for adherence to a Mediterranean diet. This observation seems to show that a score based on a theoretically defined Mediterranean diet is an effective preventive tool for measuring the of mortality and morbidity in the general population. A Mediterranean diet has been shown to have a beneficial effect on the occurrence of diseases in industrialised and non-industrialised countries. 6-8 Unfortunately, despite this worldwide promotion of the 674 BMJ 20 SEPTEMBER 2008 VOLUME 337

8 Study Knoops et al 2004 w5 Lagiou et al 2006 w7 Fung et al 2006 w8 Mitrou et al 2007 (men) w11 Mitrou et al 2007 (women) w11 Benetou et al 2008 w12 Total Reduced Relative Increased Weight (%) Relative (0.90 to 1.00) (0.77 to 1.03) (0.82 to 1.01) (0.91 to 0.97) (0.92 to 1.00) (0.81 to 0.96) (0.92 to 0.96) Fig 3 Risk of occurrence of or mortality from cancer associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size Study Scarmeas et al 2006 w9 Gao et al 2007 (men) w10 Gao et al 2007 (women) w10 Total Mediterranean diet, a progressive shift to a non- Mediterranean dietary pattern, even in countries bordering the Mediterranean sea, has progressively developed. 9 It thus seems urgent to identify an effective preventive strategy to decrease the burden related to dietary habits in the general population. Limitations Some limitations of this study can be identified. The Mediterranean diet is not a homogeneous pattern of eating, and heterogeneity on the score items exists. How to group some food categories such as legumes, nuts, and milk and dairy products; the real importance of different types of meat; and the establishment of the moderate amount of alcohol intake are still matters of dispute among researchers and can differ among the selected studies. None the less, the key characteristics of a Mediterranean diet were present in all the studies, and the overall analysis seemed not to be significantly influenced by these differences. In addition, the use of a score for estimating a dietary pattern is limited by subjectivity, conditioned by the available data and the main objectives of the study, and so possibly determining a great variability in the interpretation of the results. Finally, a further limitation exists in the different adjustment for potential confounders seen among the included studies. This difference could have determined a residual confounding within the studies, especially for the non-mediterranean cohorts. However, the sensitivity analysis according to the quality of Reduced Relative Increased Weight (%) Relative (0.70 to 0.98) (0.80 to 1.08) (0.72 to 1.00) (0.80 to 0.96) Fig 4 Risk of Parkinson s disease and Alzheimer s disease associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect size WHAT IS ALREADY KNOWN ON THIS TOPIC The Mediterranean diet is a well known model of diet for primary and secondary prevention of major chronic diseases An adherence score can be used to assess the adherence of a specific population to the rules of a traditional Mediterranean diet WHAT THIS STUDY ADDS Greater adherence to a Mediterranean diet confers a significant protection for overall mortality, as well as cardiovascular disease mortality and incidence of cancer and degenerative diseases The adherence score based on a theoretically defined Mediterranean diet could be an effective preventive tool for reducing the of mortality and morbidity in the general population the studies, which also included the presence or not of adjustment factors, showed no significant influence of residual confounding on the overall findings of our meta-analysis. Conclusions This meta-analysis shows that adherence to a Mediterranean diet can significantly decrease the of overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson s disease and Alzheimer s disease. These results are strictly concordant with current guidelines and recommendations from all the major scientific associations that strongly encourage a Mediterraneanlike dietary pattern for primary and secondary prevention of major chronic diseases. Contributors: See bmj.com. Funding: None. Competing interests: None declared. Ethical approval: Not needed. Provenance and peer review: Not commissioned; externally peer reviewed. 1 Willett WC, Sacks F, Trichopoulou A, Drescher G, Ferro-Luzzi A, Helsing E, et al. Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr 1995;61: Serra-Majem L, Roman B, Estruch R. Scientific evidence of interventions using the Mediterranean diet: a systematic review. Nutr Rev 2006;64:S KeysA.Seven countries: a multivariate analysis of death and coronary heart disease. Cambridge, MA: Harvard University Press, De Lorgeril M, Salen P, Martin J, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional factors and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon diet heart study. Circulation 1999;99: Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol 2002;13: WHO Study Group. Diet, nutrition, and the prevention of chronic diseases. Geneva: WHO, 2003: American Heart Association Nutrition Committee, Lichtenstein AH, Appel LJ, Brands M, Carnethon M, Daniels S, et al. Diet and lifestyle recommendations revision 2006: a scientific statement from the American Heart Association Nutrition Committee. Circulation 2006;114: US Department of Health and Human Services, US Department of Agriculture. Dietary guidelines for Americans thed. Washington, DC: US Government Printing Office, 2005 (available at 9 Sofi F, Vecchio S, Giuliani G, Martinelli F, Marcucci R, Gori AM, et al. Dietary habits, lifestyle and cardiovascular factors in a clinically healthy Italian population: the Florence diet is not Mediterranean. Eur J Clin Nutr 2005;59: Accepted: 6 July 2008 BMJ 20 SEPTEMBER 2008 VOLUME

9 Quality of clinical aspects of call handling at Dutch out of hours centres: cross sectional national study Hay P Derkx, 1 Jan-Joost E Rethans, 2 Arno M Muijtjens, 3 Bas H Maiburg, 4 Ron Winkens, 5 Harrie G van Rooij, 6 JAndréKnottnerus 1 EDITORIAL by Car and colleagues 1 Department of General Practice, Maastricht University, Maastricht, Netherlands 2 Skillslab, Maastricht University 3 Department of Educational Development and Research, Maastricht University 4 Department of General Practitioner Vocational Training, Maastricht University 5 Centre for Quality of Care Research, Department of General Practice, Maastricht University 6 Primary Care Out-of-hours Centre,Tilburg,Netherlands Correspondence to: H P Derkx hay.derkx@medicinfo.nl Cite this as: BMJ 2008;337:a1264 doi: /bmj.a1264 This article is an abridged version of a paper that was published on bmj.com. Cite this article as: BMJ 2008;337:a1264 ABSTRACT Objective To assess the quality of telephone triage by following the consecutive phases of its care process and the quality of the clinical questions asked about the patient s clinical condition, of the triage outcome, of the content of the home management advice, and of the safety net advice given at out of hours centres. Design Cross sectional national study using telephone incognito standardised patients. Setting The Netherlands. Participants 17 out of hours centres. Main outcome measures Percentages of clinical obligatory questions asked and items within home management and safety net advice, both in relation to preagreed standards, and of care advice given in relation to the required care advice. Results The telephone incognito standardised patients presented seven clinical cases three times each over a period of 12 months, making a total of 357 calls. The mean percentage of obligatory questions asked compared with the standard was 21%. Answers to questions about the clinical condition were not always correctly evaluated from a clinical viewpoint, either by triagists or by general practitioners. The quality of information on home management and safety net advice varied, but it was consistently poor for all cases and for all out of hours centres. Triagists achieved the appropriate triage outcome in 58% of calls. Conclusion In determining the outcome of the care process, triagists often reached a conclusion after asking a minimal number of questions. By analysing the quality of different phases within the process of telephone triage, evaluation of whether an appropriate triage outcome has been arrived at by means of good clinical reasoning or by an educated guess is possible. In terms of enhancing the overall clinical safety of telephone triage, apart from obtaining an appropriate clinical history, adequate home management and safety net advice must also be given. INTRODUCTION Most studies on the quality of telephone triage have focused on analysing the outcomes of triage. However, to assess the safety with more accuracy, the quality of the care process itself needs to be analysed. This includes assessment of the clinical quality of questions asked, as well as evaluation of the answers and the care advice given. The objective of this study was to make a global assessment of quality, specifically the quality of the clinical questions asked, the evaluation of the answers, the triage outcomes, and the content of the home management and safety net advice, in telephone consultations carried out by triagists at out of hours centres in the Netherlands. METHODS To assess the quality of clinical case handling at out of hours centres, telephone incognito standardised patients presented seven different clinical cases three times each to 17 different out of hours centres over a period of 12 months, totalling 357 calls. Cases, protocols, and scenarios The cases were to be based on symptoms that are known to present frequently on the telephone to out of hours centres, 12 and the care advice given as an outcome of the telephone consultation needed to be straightforward. In the Netherlands the triagist can select different triage outcomes depending on the degree of urgency. For this study we decided for practical reasons that no case would need an outcome of immediate care or a home visit. The cases were designed to show the importance of accurate history taking. Therefore, six of the seven cases to be presented consisted of three pairs of almost identical cases. For each of these parallel pairs of cases only one answer to an obligatory question differed (the discriminating answer). The box shows the obligatory questions for case 1; see bmj.com for the remaining cases. The table describes the discriminating answers for each parallel pair of cases. Protocols for telephone triage include questions about possible causes and the consequences of the symptom presented, questions to compensate for the lack of visual information, and advice about home management care to be given if the triage outcome does not include a face to face consultation for example, type and dose of analgesic drug. Finally, they include advice about the circumstances in which the patient should call back ( safety net advice ). 3-5 Discriminating answer for each parallel pair of cases Case Age Clinical problem Discriminating answer 1 5 years Fever Not applicable 2 Adult Nosebleed None 3 Adult Nosebleed Noticed several bruises in past few hours 4 Adult Fever None 5 Adult Fever Irregular usage of antimalarial drugs in past two weeks 6 5 years Vomiting None 7 5 years Vomiting Had head injury a few hours ago 676 BMJ 20 SEPTEMBER 2008 VOLUME 337

10 Obligatory questions to be asked for case 1, a 5 year old child with fever Can you describe the child s behaviournow? How high is the temperature of the child? Has the child had a fit? Does the child have pain anywhere? Has the child got, or has he or she had, a headache? Can the child touch forehead on knees (or kiss knees)? Does the child seem breathless or is there indrawing of the chest/tummy? How much fluid has been taken in the past 12 hours? When did the child last pass urine? Does the child have a rash? Did the child recently travel abroad? We presented these protocols to a panel of general practitioners with experience in telephone triage at out of hours centres. They agreed on the obligatory questions that should be asked about the different clinical symptoms and what should be discussed with the patient in relation to home management and safety net advice. Telephone incognito standardised patients Lay people have been used as unannounced or incognito standardised patients to assess the performance of students and healthcare professionals during face to face and telephone consultations. Evidence shows that this method is a reliable and valid instrument to assess clinical performance. 6-9 Trained telephone incognito standardised patients presented the cases. They received a validated scenario, with information on their personal data, the clinical problem, their personal situation (such as feeling anxious), and the answers to be given to the obligatory questions if they were asked. If they were invited to come for a consultation at the out of hours centre, they were instructed to ask for the reason for this request. Out of hours centres We asked all 105 out of hours centres in the Netherlands for permission to be selected for the research study. Of the 98 centres that gave permission, we randomly selected 17 (but did not inform them) on the basis of the size of the 12 provinces of the Netherlands. After the study we asked the 17 out of hours centres by letter whether they had detected a telephone incognito standardised patient during the previous 12 months. Assessment and analysis The required frequency of presentation of the cases developed was based on generalisability theory (see bmj.com). In a first batch, the seven cases were presented five times to five of the 17 selected out of hours centres. The remaining 12 out of hours centres were called three times for the same seven cases. We made a transcript of every recorded call and assigned each call to two medical students as raters. They scored calls independently by using the standard protocol as a checklist. They rated each item by marking Yes (=1: question is asked/advice is given) or No (=0: question is not asked/advice is not given). Obligatory questions (%) Home management and safety net advice (%) Case Out of hours centre Box plots (distributions) of percentage obligatory questions asked versus standard and percentage home management and safety advice provided versus standard. Left panel shows distribution for all calls (n=357), middle panel shows distributions per case (51 calls perboxplot), andright panelshowsdistributions per out ofhourscentre (21 callsper box plot). Eachboxplotshowsmeanvalue (solid line), 25th and 75th centiles (lower and higher boundaries of grey box) indicating central part, and 10th and the 90th centiles (grey whiskers outside box) indicating outer parts of distribution BMJ 20 SEPTEMBER 2008 VOLUME

11 WHAT IS ALREADY KNOWN ON THIS TOPIC Research on the quality of telephone triage is often focused on the quality of the outcomes Little is known about the quality of the different phases of the care process of telephone triage by triagists WHAT THIS STUDY ADDS Care advice was given after asking (too) few obligatory questions Answers to those questions, if asked, were not always interpreted correctly by the triagist or general practitioner The quality of home management and safety net advice was low For each call, we recorded the following variables as indicators of quality: percentage of obligatory questions asked in relation to the agreed standard set of questions; percentage of items within home management and safety net advice in relation to the agreed standard set of items; percentage of obligatory questions asked in relation to all questions asked; percentage of appropriate care advice given in relation to the required care advice. For each of these variables we used the average percentage of the two raters for further analysis of each case for all out of hours centres and for each out of hours centre for all cases. RESULTS For 58% of all calls the required urgency level advised was as set by the scenario panel. Urgency was underestimated in 41% of calls and overestimated in 1% of calls. The overall mean percentage of obligatory questions asked was equal to 21% of the standard (figure). The mean percentage varied between 15% at out of hours centre number 6 and 42% at number 5. We found a similar pattern between cases: some variation between cases existed, but for all cases the mean percentage was far below the standard. The overall mean for the content of home management and safety net advice was 40% (figure). The variation between cases was from 26% for the case of an adult with fever to 56% for the case of an adult with a nosebleed. On average, 54% of all questions asked were obligatory questions. This percentage varied from 32% to 73% between cases and from 46% to 65% between out of hours centres. The figure shows the difference in performance between individual triagists at the same out of hours centre. Some managed a score of almost 90% for relevant home management and safety net advice, and others gave none at all. The same triagist handled the same case twice in 3% of the calls; all other calls were handled by different triagists. The triagists referred 2% of the telephone incognito standardised patients to the out of hours centre in the region of their permanent residence without any triage. None of the 17 out of hours centres indicated that they had detected any call made by a telephone incognito standardised patient. For 153 calls the required care advice resulting from the telephone triage was to come to the out of hours centre. This care advice was given on 17 occasions, on seven of which no clinical reason for coming could be identified. DISCUSSION The results of this study identify shortcomings and educational needs in telephone triage, indicated by assessing the quality of phase one (asking appropriate questions and evaluating the answers) and phase two (giving care advice) of the process of telephone triage. The mean score for obligatory questions asked was 21% of the agreed standard. Answers to obligatory questions were not always evaluated clinically correctly. The appropriate triage outcome was reached in only 58% of calls. Strength and weaknesses Triagists are known to respond very accurately to cases of high urgency but not to moderate and low urgency cases such as those presented in this study. 10 Our study shows underestimation of urgency in 41% of calls and overestimation of urgency in 1%, which is in line with other results As none of the 17 out of hours centres had detected any of the telephone incognito standardised patients and different triagists handled 97% of all calls, we conclude that this study reflects the day to day performance of call handling at those centres. Cases, protocols, and scenarios In our study many obligatory questions remained unasked, and triagists often asked questions that the protocol and scenario panels did not consider necessary to determine an urgency level. Research shows that triagists often ask few questions and do not use or follow protocols. The safety of any care process is likely to be improved by following protocols, but little is known about what extent of improvement might occur. Conclusion Triagists seemed to carry out a rapid clinical scan before they came to a conclusion, without considering in sufficient detail different causes for a symptom or its possible consequences. Even when the triagist gave the required self care advice, for nearly all calls this choice could be considered to be the correct choice only by means of an educated guess as so few obligatory questions were asked. This study also shows that triagists should ideally recommend a low urgency level of care only after thorough history taking. The quality of information on home management and safety net advice varied to a large extent and was consistently low for each case and each out of hours centre. The safety of telephone triage might be enhanced by using computer based decision support systems, but further research is needed. We thank Ron Hoogenboom for helping with the statistical analysis of all data. Contributors: Each author participated as a member of the research team during each stage of this study and in writing the article. J-JER is the guarantor. 678 BMJ 20 SEPTEMBER 2008 VOLUME 337

12 Funding: A contribution was received from the Dutch Health Insurance Companies CZ (Tilburg) and VGZ (Nijmegen). The authors work was independent of the funders (the funding source had no involvement). Competing interests: None declared. Ethical approval: Not needed. Provenance and peer review: Not commissioned; externally peer reviewed. 1 Post J. Grootschalige huisartsenzorg buiten kantooruren [Large scale primary care service during after-hours] [dissertation]. Groningen, Netherlands: University of Groningen, Dale J, Williams S, Crouch R,Patel A.A study of out-of-hourstelephone advice from an A&E department. Br J Nurs 1997;6: Derkx HP, Rethans J-JE, Knottnerus JA, Ram P. Assessing communication skills of clinical call handlers working at an out of hours centre: the development of the RICE rating scale. Br J Gen Pract 2007;57: NeighbourR.The inner consultation: how to develop an effective and intuitive consulting style. Lancaster, UK: M T Press, Higgs RH, Armstrong E, Blache G, Jessopp L, Peters M. Telephone consultations [electronic response to Car J, Sheikh A. Telephone consultations]. BMJ Rethans J-J, Gorter S, Bokken L, Morrison L. Unannounced standardized patients in real practice: a systematic literature review. Med Educ 2007;41: Van der Vleuten C, Swanson D. Assessment of clinical skills with standardized patients: state of the art. Teach Learn Med 1990;2: Rethans JJ, Norcini JJ, Baron-Maldonado M, Blackmore D, Jolly BC, LaDuca T, et al. The relationship between competence and performance: implications for assessing practice performance. Med Educ 2002;36: Moriarty H, McLeod D, Dowell A. Mystery shopping in health service evaluation. Br J Gen Pract 2003;53: Leprohon J, Patel VL. Decision-making strategies for telephone triage in emergency medical services. Med Decis Making 1995;15: Wahlberg AC, Wredling R. Telephone nursing: calls and caller satisfaction. Int J Nurs Pract 1999;5: O Brien RP, Miller TL. Urgent care center pediatric telephone advice. Am J Emerg Med 1990;8: Lattimer V, George S, Thompson F, Thomas E, Mullee M, Turnbull J, et al. Safety and effectiveness of nurse telephone consultation in out of hours primary care: randomised controlled trial. BMJ 1998;317: Crouch R. An investigation into effects of a computer based decision support program on accident and emergency nurses assessment strategies in telephone consultation [dissertation]. Brighton, UK: University of Surrey, Dale J, Williams S, Foster T, Higgins J, Snooks H, Crouch R, et al. Safety of telephone consultation for non-serious emergency ambulance service patients. Qual Saf Health Care 2004;13: Accepted: 6 July 2008 Misperceptions and misuse of Bear Brand coffee creamer as infant food: national cross sectional survey of consumers and paediatricians in Laos Hubert Barennes, 1 Todisoa Andriatahina, 1 Vattanaphone Latthaphasavang, 1 Margot Anderson, 2 Leila M Srour 2 1 Institut de la Francophonie pour la Médecine Tropicale, BP 9519, Vientiane, Lao Popular Democratic Republic 2 Health Frontiers, University of Health Sciences, Vientiane Correspondence to: H Barennes hubert.barennes@auf.org Cite this as: BMJ 2008;337:a1379 doi: /bmj.a1379 This article is an abridged version of a paper that was published on bmj.com. Cite this article as: BMJ 2008;337:a1379 ABSTRACT Objective To investigate the use of Bear Brand coffee creamer as a food for infants and the impact on consumers of the logo of a cartoon baby bear held by its mother in the breastfeeding position. Design Interviews with paediatricians throughout the country and a national survey of potential consumers regarding their perceptions and use of the Bear Brand coffee creamer. Setting 84 randomised villages in south, central, and northern Laos. Participants 26 Lao paediatricians and 1098 adults in households in a cluster sampling. Results Of the 26 paediatricians, 24 said that parents often or sometimes fed this product to infants as a substitute for breast milk. In the capital city, paediatricians said that mothers used the product when they returned to work. In the countryside, they reported that poor families used it when the mother was ill or had died. Of 1098 adults surveyed, 96% believed that the can contains milk; 46% believed the Bear Brand logo indicates that the product is formulated for feeding to infants or to replace breast milk; 80% had not read the written warning on the can; and over 18% reported giving the product to their infant at a mean age of 4.7 months (95% confidence interval 4.1 to 5.3). Conclusion The Bear Brand coffee creamer is used as a breast milk substitute in Laos. The cartoon logo influences people s perception of the product that belies the written warning This product is not to be used as a breast milk substitute. Use of this logo on coffee creamer is misleading to the local population and places the health of infants at. INTRODUCTION The international code of marketing of breast milk substitutes is intended as a minimum requirement by all governments and aims to protect infant health by preventing inappropriate marketing of breast milk substitutes. 1 Product labels are often the only source of information available to consumers regarding the content and uses of a product, and public interest groups have documented multiple instances of formula companies providing misleading information on infant formula labels. 2 In Laos a popular coffee creamer is marketed with an illustration of a mother bear holding a baby bear in the breastfeeding position (figure). The manufacturing company uses the same Bear Brand logo on its canned sterilised cows milk product and on infant formula products for infants from 6 months. A warning on the can states This product is not to be used as breast milk substitute in English, Thai, and Lao. There is also an illustration of a feeding bottle with a cross through it. We have encountered infants and children admitted to hospital with protein calorie malnutrition who were BMJ 20 SEPTEMBER 2008 VOLUME

13 Label from Bear Brand coffee creamer fed this product exclusively. We conducted two surveys to determine whether or not the practice of feeding this coffee creamer to infants is widespread in Laos. We also explored the impact of the logo on people s perception of the product s appropriateness for infants. METHODS Laos is one of the poorest countries in southeast Asia. It is a multiethnic and multilingual country with more than 45 languages spoken. Literacy rates range from 54% in rural areas to 89% in urban areas. 3 There is a high prevalence of stunting (41%) and wasting (15%) among children (0-5 years). 45 In Laos, 95% of the mothers breast feed their newborn children but the rate of exclusive breast feeding is below 28%. 6 Inadequate breast feeding and weaning practices contribute to high rates of malnutrition and infant and child mortality. 78 One author (LMS) interviewed 26 paediatricians in eight of the 17 provinces to collect information on parents use of the Bear Brand coffee creamer as a substitute for breast milk. In 2007 we conducted a sample survey in Laos in five representative provinces. We randomly selected and enrolled 84 villages and an average of 14 households with one adult aged over 18 from each. A semistructured questionnaire examined knowledge and use of the coffee creamer with the Bear Brand logo. Participants gave their informed consent to participate in the survey. Table 1 Responses of 1023 adultswhenasked What does this logo mean? (respondents shown Bear Brand logo, see figure) Response No (%) Product is good for infants 402 (39.3) Product is good for children 115 (11.2) Advertising 110 (10.7) Product is replacement for breast milk 66 (6.5) Product is for adults 13 (1.3) Product is made from animals 6 (0.6) Bears love their children 4 (0.4) Product is for everyone 3 (0.3) Not sure 242 (23.7) Other response 2 (0.2) We calculated that we needed a sample size of 983 people, based on an estimated use of coffee creamer of 20%, a 4% precision with α=0.05, and 90% power, and we added 10% more people to account for incomplete data, resulting in a sample of RESULTS Of the 26 paediatricians interviewed, 13 reported that parents often feed the Bear Brand coffee creamer to infants as a substitute for breast milk. Eleven reported that parents sometimes feed the product to infants. Paediatricians in Vientiane reported that mothers use the product when they return to work. In the countryside poor families use the product as a breast milk substitute when the mother is ill or dies. Paediatricians have encountered infants and children admitted to hospital with protein calorie malnutrition who had been fed this product exclusively and often reported similar stories (see bmj.com). Of 1098 respondents, 570 (51%) lived in remote rural area, 364 (33%) lived in a semirural area, and 164 (15%) lived in an urban area. Of the adults interviewed, 1031/ 1098 (94%) recognised the can; 994/1031 (96%) believed that it contained milk; and only 21/1031 (2%) identified the contents correctly as coffee creamer. In total, 191 (19%) reported giving the coffee creamer with the Bear Brand logo to infants at a mean age of 4. 7 months (95% confidence interval 4.1 to 5.3). The main reasons given were that they thought it complemented breast feeding (40%), was good for infants growth (19%), was a substitute for breast milk (17%), and was cheap (11%). Of 1031 people, 824 (80%) said they had not read the text warning on the can. Tables 1 and 2 describe the respondents understanding of the Bear Brand logo and the feeding bottle with a cross through it. DISCUSSION Bear Brand coffee creamer is a recognised and well distributed product in diverse parts of Laos. Nearly half of surveyed adults believe that the cartoon logo on the can means that the product is good for infants or a replacement for breast milk. Nearly a fifth of parents had given the product to their young infants. 680 BMJ 20 SEPTEMBER 2008 VOLUME 337